cross-posted from: https://sh.itjust.works/post/5572424
This might have been discussed to death by now, unfortunately I couldn’t find any discussion on it on Lemmy. Though I would love to be corrected on that!
How does an always on incognito Chromium with uBlock Origin on medium mode (and other hardening/privacy settings enabled) compare to Brave (with e.g. Privacy Guides’ recommended settings) with respect to security and privacy on Linux[1]?
Commonly heard whataboutisms:
- “With the looming advent of Manifest v3, this discussion might not be very relevant for long.” I’m aware.
- “Just use Firefox/Librewolf or any other privacy-conscious browser that isn’t Chromium-based.” I already do, but some websites/platforms don’t play nice on non-Chromium-based browsers due to Google’s monopoly on the web. Sometimes I can afford to not use that website/platform, but unfortunately not always.
- “Brave’s [insert controversy] makes them unreliable to take services from.” Honestly, I think that if both solutions are as effective that a reason like this might be sufficient to tip the balance in favor of one. Because ultimately this all comes down to trust.
- "Just use Ungoogled Chromium." Some more knowledgeable people than me advice against it. Though, I’d say I’m open to hear different opinions on this as long as they’re somewhat sophisticated.
- “Just use [insert another Chromium-based browser].” If it has merits beyond Brave and Chromium with respect to security and privacy, I’ll consider it.
Thanks in advance!
- I can be more specific about which distro I prefer using, but I don’t think it matters. I might be wrong though*.
Yes, I could say come to arch but you seem happy in fedora 😉
Hehe :P . True dat. Maybe one day ;) . Perhaps I’ll just spin up a distrobox in order to get access to Brave through the AUR, but this (excellent) article has worsened my already bad paranoia to clearly unhealthy levels 🤣. So, it seems out of question for now 😅. Though I might be able to spin it up in a Wolfi container. Pessimism doesn’t help though 🤣.
Man you’ve gone down a security worm hole that makes me wonder if you should really be running qubes-OS rather than Fedora 🤣.
Seriously if you need more than the chromium sandbox for brave and want simplicity just use firejail.
The article you linked to is a wonderfully detailed write up but it is more geared towards those using containers that will be providing services (web, sql, etc) if you just want a browser in a secure container then any of the implementations will be fine for you. The browser is not a vector used to gain access to your OS directly but what you download potentially is so with that in mind your downloads folder should really be a CLAMFS folder or a target folder for on-access scanning by clamav.
Hahaha 🤣. Honestly I would, if my device could handle.
Madaidan strikes (yet) again. F*ck my paranoia…
Very interesting insights! Thank you so much! Would you happen to know of resources that I might refer to for this?
Question: Why do you think need such high security for a browser?
Clam av on access scan: https://wiki.archlinux.org/title/ClamAV#OnAccessScan
ClamFS: https://github.com/burghardt/clamfs
Your help is much appreciated!
Good prompt! I actually started questioning my own motivations from this. And I’d say that the best I could come up with was that it’s required in order to attain the “peace of mind” from having properly secured my browser activity; which happens to be the primary activity on my device anyways.
Valid response, but why do you need to protect the OS from the browser when the browser (Brave) is already sandboxing and the browser is not an attack vector that can be directly exploited to gain access/root on your OS?
What I mean is that the tabs themselves are sandboxed to protect accounts that are opened in each from being breached, the bowser itself is obfuscating your fingerprint and blocking known bad actor sites etc so this leaves only what you manually download and here the browser will warn you if a given download has the potential to harm.
So unless you are downloading files from very questionable locations I can’t see the need for a containerised browser.
Containers are good and yes have flaws but the main purpose of them is to add another layer between the application and the OS so if application is exploited the attacker has to break another wall/layer to get to the real root.
I know in April 2021 the was a PoC that used JavaScript to reverse the effect of a patch which allowed an attacker to break out of the chromium sandbox, but that was never used and if it was the attacker would first need to breach a site to deploy the code that you would then execute by visiting the site or it would be fed to you via a phishing attempt. Both of these delivery methods would need to be very stealthy and fast. currently there are 4 known CVEs for brave: (sorry for long link)
https://www.cvedetails.com/vulnerability-list.php?vendor_id=16266&product_id=36540&version_id=0&page=1&hasexp=0&opdos=0&opec=0&opov=0&opcsrf=0&opgpriv=0&opsqli=0&opxss=0&opdirt=0&opmemc=0&ophttprs=0&opbyp=0&opfileinc=0&opginf=0&cvssscoremin=0&cvssscoremax=0&year=0&cweid=0&order=1&trc=3&sha=74c1df28c6d85bd121726a90109559ec94ea3549
None of these provide an attack vector that will allow access.
I’ve been enjoying your responses a lot! I just wanted to express my gratitude one more time!
Uhmm…, but I think that somewhat of a misunderstanding might have happened somewhere.
Just to be clear. I acknowledge Brave’s (or rather Chromium’s for that matter) sandbox capabilities. I’m not necessarily afraid of whatever I’m doing inside to break out of the sandbox. Sure, the ‘risk’ (if at all) can be further circumvented with the use of VMs and whatnot and for some people this approach is justified. But me lamenting on using something like Qubes (eventually) is more about having an OS that actually has sane security defaults. And having browsers run in VMs is just part of that. Currently, I just want a secure and private browser to use on desktop. So far, it seems that Brave is superior over Chromium due to added features like fingerprint-spoofing, the inevitable discontinuation of Manifest v2 etc.
What I am afraid of is how secure (continued) operation within containers would be. So even if Brave (or whichever browser for that matter) is not the culprit, the rest of the container environment might endanger the rest of my system. Of course, I’m a total noob so I might be talkin’ outta my A$$. So please correct me if my understanding is faulty.
Hehe, I guess if I would be forced to do a thing like that I would do so within a VM 😅.
So I’ve mostly been using well-integrated ‘pet-containers’ like the ones known from Distrobox (with a relevant recent feature). Aside from those I’ve been exposed to the earlier article and to this video. These ‘expositions’ have made me go from a Distrobox-enjoyer to a pessimist that doesn’t dare to come close to them until I’ve better educated myself on them 🤣.
Thanks man, means a lot these days.
If your container for brave is running but the browser itself is closed, there is no way for to happen within the container because the software that would be connected to the internet is closed/quit/stopped. In fact that container should be reported as down by whichever management subsystem is provided by said container (portainer, lxd, systemd-namespaces, etc)
I think you should look more into what containers are and can do, You previously said that your system is low power but distrobox is making loads of of full OS/distro containers which for the most part act like a VM. Distrobox is a good way to test drive a distro OR allow a dev to ensure the app they’ve made works on their target distro’s for chosen use case.
All you really need to do is run a single application within a container, not a whole distro!/os Why do I say this? Well resource consumption for one and why replicate an entire distro/os when an app can be run inside a container: https://bacchi.org/posts/brave-in-docker/
Additionally I spoke about attack vectors, running another distro/OS inside a docker may well have samba, ssh running by default, If the container for that is not firewalled that is is an attack vector that will allow RCE and exploits be run inside that container!
The first minute of that video talks of nginx webserver image, That is a webserver running inside a container, with distrobox you have the rest of the OS inside the container as well as nginx. Do you get what I say now?
I suggest you use the above link I gave to look into running just a browser within a container, drop distrobox (unless you need to test drive distros) and learn about running a single application within a container, when you can do that find a container framework that provides the security you want/like then run your “untrusted” applications in containers and rejoice with a slightly faster machine.
EDIT: Additionally wolfi is based on Alpine, This is a popular server distro, If you want to install wolfi you’ll need to know how to install alpine, which is similar to installing gentoo as it uses bootstrap images, don’t be surprised if the desktop experience is a bit …erm lacking as that is not the focus of alpine or wolfi ! Good luck